Add Azure Event Hubs persistence functional test #7477
Conversation
| var resourceEvent = await rns.WaitForResourceAsync("resource", e => e.Snapshot.State?.Text == KnownResourceStates.Running, cts.Token); | ||
| var ehContainerId1 = resourceEvent.Snapshot.Properties.FirstOrDefault(x => x.Name == "container.id")?.Value?.ToString(); | ||
|
|
||
| resourceEvent = await rns.WaitForResourceAsync("resource-storage", e => e.Snapshot.State?.Text == KnownResourceStates.Running, cts.Token); | ||
| var storageContainerId1 = resourceEvent.Snapshot.Properties.FirstOrDefault(x => x.Name == "container.id")?.Value?.ToString(); |
There was a problem hiding this comment.
(nit) it may be more readable if you refactor this code out into a helper method "GetContainerId(string resourceName)`.
| await app.StopAsync(cts.Token).WaitAsync(TimeSpan.FromMinutes(1)); | ||
|
|
||
| using var builder2 = TestDistributedApplicationBuilder.Create(testOutputHelper); |
There was a problem hiding this comment.
This test could use this new functionality:
That way we knew the app was stopped.
There was a problem hiding this comment.
It's on by default. But I decided to disable it on purpose, it's not required for the test (we check we don't have a different container, the state of the previous one doesn't matter) and not waiting makes the test faster (19s -> 12s)
|
|
||
| // NOTE: Waiting for Running state doesn't guaranty that the container-id property is set (for this resource at least) | ||
|
|
||
| resourceEvent = await rns.WaitForResourceAsync("resource-storage", e => GetContainerId(e) is not null, cts.Token); |
There was a problem hiding this comment.
I had some problems locally with this. I am trying to read the "container-id" property of the snapshot. However if I wait for the Running state and then read the id right after it returns null most of the time. For the previous resource (a parent) it works all the time. So now I use the actual value as the predicate. Also these containers were already running on my machine (persistent from a previous run) so the container always has an id and is running.
There was a problem hiding this comment.
It is a little surprising that a container in Running state does not have a container ID set since on DCP side this is done as one change that should be atomic. But what you are doing here certainly makes sense.
The other thing to keep in mind is that even for running containers, the sequence of events represents the lifetime of the Container object in DCP. So the first event will always be a "Created" event with no container ID, and then DCP will notice that the existing Docker container can be reused, so it will fill the ContainerID property and change its Container object state to Running. And that change will be delivered as a separate event.
| var resourceEvent = await rns.WaitForResourceAsync("resource", e => e.Snapshot.HealthStatus == HealthStatus.Healthy, cts.Token); | ||
| var ehContainerId1 = GetContainerId(resourceEvent); | ||
|
|
||
| // NOTE: Waiting for Running state doesn't guaranty that the container-id property is set (for this resource at least) |
There was a problem hiding this comment.
| // NOTE: Waiting for Running state doesn't guaranty that the container-id property is set (for this resource at least) | |
| // NOTE: Waiting for Running state doesn't guarantee that the container-id property is set (for this resource at least) |
Description
This is the first functional test verifying that persistence is effective for a container. I might have missed more efficient ways to do this. Currently only checking that the same container is used when a different app is started. Once approved we can add the same thing to other containers.
Checklist
<remarks />and<code />elements on your triple slash comments?breaking-changetemplate):doc-ideatemplate):